Table of contents
€140 billion in cloud and software overcharges per year for European organizations, and one in two CIOs unable to measure the benefits of bundled AI: this is the finding published by Cigref on May 28, 2026, study conducted by Asteres. The report, relayed by Le Monde Informatique, quantifies for the first time the macroeconomic effect of rising cloud and software prices on the Union: European organizations' cloud and software spending has experienced an average annual inflation of 8.7% over the past three years, with a projection of 12% per year for the 2026-2030 period. The authors also estimate the loss of added value at €107 billion per year, of which €93 billion exits the European economy. The survey is based on a panel of 54 IT decision-makers from European companies and administrations.
A massive reorientation of capex towards AI, facing a marginal European fabric
The pricing trajectory documented by Asteres is part of an unprecedented industrial investment cycle. According to guidance published by AWS, Azure, Google, Meta, and Oracle in April-May 2026, the consolidated envelope of the five major hyperscalers exceeds $600 billion in 2026, up sharply from the roughly $380 billion committed in 2025 - an increase of around 55% - with approximately 75% directed towards AI infrastructure, an investment that mechanically affects downstream pricing balances. In contrast, the European market remains marginal: European providers captured only ~15% of the EU cloud market in 2022, according to the ECIPE (European Centre for International Political Economy). This situation is not new for IT departments: CIO Online noted it as early as December 2025, placing vendor dependency and double-digit cloud and software inflation at the top of CIOs' concerns for 2026. However, an alternative signal is emerging: the OVHcloud, DEEP, and Clever Cloud consortium has been selected by the European Commission for a sovereign cloud contract worth €180 million over 6 years, notified on April 17, 2026 - a scale that remains incomparable to the aggregated spending base of member states, but which opens a first line of European public procurement in this area. ActuIA was already documenting the product trajectory of the French actor with the launch of AI Endpoints, its serverless open-source AI platform.
What European CIOs identify as drivers of increases
Surveyed about the mechanisms that allow them to absorb increases, respondents to the Asteres panel point first to technological or contractual lock-in - vendor lock-in - cited by 40% of them. Next, and this is the most recent shift, is the native integration of AI functions supposed to justify the surcharges, cited by 32% of respondents, which now surpasses programmed obsolescence techniques (30%). The constrained nature of bundled AI is numerically documented: 21% of surveyed CIOs were imposed a natively integrated AI option in their solutions, with a surcharge, without the possibility of refusing the option. The benefit counterpart remains difficult to objectify - one in two respondents fails to measure AI benefits or believes they will only be achievable at the cost of a reorganization. Recent academic literature confirms this scale gap: an RCT study published on arXiv (2504.11443), conducted on nearly 6,000 workers within 56 companies with Microsoft 365 Copilot, measures gains at the task level - emails processed about 7% faster, documents completed 12% more quickly - without isolating a measurable return at the organizational level. The gains are verified at the workstation level but do not yet translate, according to the study authors, into a measurable effect on the budget line for the 56 companies in the sample.
The four pricing increase mechanisms identified by European CIOs (Asteres/Cigref study, 54 European IT decision-makers)
| Mechanism | Concerned CIOs | Note |
|---|---|---|
| Contractual lock-in (vendor lock-in) | 40 % | Historical mechanism, still dominant |
| Native AI integrated by design | 32 % | 2nd driver, ahead of planned obsolescence |
| Planned obsolescence | 30 % | Classic technique, now surpassed by AI |
| Imposed AI option without refusal possibility | 21 % | Non-negotiable surcharge, no contractual opt-out |
Source: Asteres study for Cigref, May 2026 - sample of 54 IT decision-makers, statistical representativeness limit to be considered.
A FinOps asymmetry that classic levers fail to correct
On the budget management tools side, the gap between classic cloud and AI charges remains poorly documented publicly. Due to a lack of consolidated source at this stage, several FinOps practitioners estimate that negotiable discounts - saving plans and volume commitments - cap around 20% on generative AI charges, compared to 45 to 55% usually obtained on classic cloud resources: an editorial inference order of magnitude, to be confirmed as soon as an independent barometer measures it. The sovereign path also does not solve the price equation: the official launch of the SAP Sovereign Cloud in France via Bleu, an Orange-Capgemini joint venture, on March 19, 2026, targets OIV and public administration; effective pricing conditions have not been made public at this stage. As for the Microsoft-OpenAI amendment of April 2026, which ends Azure's exclusivity on OpenAI workloads and allows multi-cloud deployment of APIs, its effect is measured on a narrow scope: at first analysis, the renegotiation lever applies to tokenized API lines; the per-seat M365 Copilot and Workspace bundles, sold as an integrated seat option, remain out of reach of this opening. On the European organizations' side, the adaptation behaviors declared to the Asteres panel illustrate the budgetary arbitration described by Cigref (n=54): to absorb overcharges, according to the Asteres survey, about one in two respondents reduces their other digital expenses; a third increases their IT budget. Cigref qualifies these arbitrations with a dry formula: "local ESN, recruitments in Europe, R&D, and productive investments then serve as adjustment variables."
FinOps AI figures: no consolidated public barometer at this stage
Comparisons of savings between AI saving plans and classic cloud reservations circulate in FinOps circles and the specialized press, but they rely on practitioner feedback, not on a published independent study. Before using these ratios in a budget meeting or tender, check the specific contractual conditions of your provider: optimization margins vary greatly depending on the type of workload and the level of commitment.